Search: “inference”

Safety Training Modulates Harmful Misalignment Under On-Policy RL, But Direction Depends on Environment Design

arXiv:2604.12500v1 Announce Type: new Abstract: Specification gaming under Reinforcement Learning (RL) is known to cause LLMs to develop sycophantic, manipulative, or deceptive behavior, yet the condi

safetyarxiv-cs-lg

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Research

SceneCritic: A Symbolic Evaluator for 3D Indoor Scene Synthesis

DGX agent

arXiv:2604.13035v1 Announce Type: cross Abstract: Large Language Models (LLMs) and Vision-Language Models (VLMs) increasingly generate indoor scenes through intermediate structures such as layouts and

researcharxiv-cs-cl

15 Apr 2026

Simulation as Supervision: Mechanistic Pretraining for Scientific Discovery

arXiv:2507.08977v4 Announce Type: replace-cross Abstract: Scientific modeling faces a tradeoff between the interpretability of mechanistic theory and the predictive power of machine learning. While ex

SinkSAM-Net: Knowledge-Driven Self-Supervised Sinkhole Segmentation Using Topographic Priors and Segment Anything Model

model-releasesarxiv-cs-cv

arXiv:2410.01473v2 Announce Type: replace Abstract: Soil sinkholes significantly influence soil degradation, infrastructure vulnerability, and landscape evolution. However, their irregular shapes, com

SOAR: Self-Correction for Optimal Alignment and Refinement in Diffusion Models

arXiv:2604.12617v1 Announce Type: cross Abstract: The post-training pipeline for diffusion models currently has two stages: supervised fine-tuning (SFT) on curated data and reinforcement learning (RL)

TCL: Enabling Fast and Efficient Cross-Hardware Tensor Program Optimization via Continual Learning

model-releasesarxiv-cs-lg

arXiv:2604.12891v1 Announce Type: new Abstract: Deep learning (DL) compilers rely on cost models and auto-tuning to optimize tensor programs for target hardware. However, existing approaches depend on

The Non-Optimality of Scientific Knowledge: Path Dependence, Lock-In, and The Local Minimum Trap

arXiv:2604.11828v1 Announce Type: new Abstract: Science is widely regarded as humanity's most reliable method for uncovering truths about the natural world. Yet the trajectory of scientific discovery

Think Through Uncertainty: Improving Long-Form Generation Factuality via Reasoning Calibration

arXiv:2604.12046v1 Announce Type: new Abstract: Large language models (LLMs) often hallucinate in long-form generation. Existing approaches mainly improve factuality through post-hoc revision or reinf

researcharxiv-cs-cl

Time-reversed Flow Matching with Worst Transport in High-dimensional Latent Space for Image Anomaly Detection

arXiv:2508.05461v3 Announce Type: replace Abstract: Likelihood-based deep generative models have been widely investigated for Image Anomaly Detection (IAD), particularly Normalizing Flows, yet their s

Tutorials

TimeSAF: Towards LLM-Guided Semantic Asynchronous Fusion for Time Series Forecasting

arXiv:2604.12648v1 Announce Type: cross Abstract: Despite the recent success of large language models (LLMs) in time-series forecasting, most existing methods still adopt a Deep Synchronous Fusion str

tutorialsarxiv-cs-ai

Toward Efficient and Robust Behavior Models for Multi-Agent Driving Simulation

arXiv:2512.05812v5 Announce Type: replace-cross Abstract: Scalable multi-agent driving simulation requires behavior models that are both realistic and computationally efficient. We address this by opt

Towards Realistic and Consistent Orbital Video Generation via 3D Foundation Priors

arXiv:2604.12309v1 Announce Type: new Abstract: We present a novel method for generating geometrically realistic and consistent orbital videos from a single image of an object. Existing video generati

ToxiTrace: Gradient-Aligned Training for Explainable Chinese Toxicity Detection

model-releasesarxiv-cs-cl

arXiv:2604.12321v1 Announce Type: new Abstract: Existing Chinese toxic content detection methods mainly target sentence-level classification but often fail to provide readable and contiguous toxic evi

Training single-electron and single-photon stochastic physical neural networks

arXiv:2604.10861v1 Announce Type: cross Abstract: The computational demands of deep learning motivate the investigation of alternative approaches to computation. One alternative is physical neural net

Tutorials

Ultra-low-light computer vision using trained photon correlations

arXiv:2604.11993v1 Announce Type: new Abstract: Illumination using correlated photon sources has been established as an approach to allowing high-fidelity images to be reconstructed from noisy camera

tutorialsarxiv-cs-cv

Uncertainty Quantification in CNN Through the Bootstrap of Convex Neural Networks

arXiv:2604.11833v1 Announce Type: new Abstract: Despite the popularity of Convolutional Neural Networks (CNN), the problem of uncertainty quantification (UQ) of CNN has been largely overlooked. Lack o

Variational Autoencoding Discrete Diffusion with Enhanced Dimensional Correlations Modeling

arXiv:2505.17384v2 Announce Type: replace-cross Abstract: Discrete diffusion models have recently shown great promise for modeling complex discrete data, with masked diffusion models (MDMs) offering a

Hardware

VFA: Relieving Vector Operations in Flash Attention with Global Maximum Pre-computation

arXiv:2604.12798v1 Announce Type: cross Abstract: FlashAttention-style online softmax enables exact attention computation with linear memory by streaming score tiles through on-chip memory and maintai

hardwarearxiv-cs-ai

VidTAG: Temporally Aligned Video to GPS Geolocalization with Denoising Sequence Prediction at a Global Scale

arXiv:2604.12159v1 Announce Type: new Abstract: The task of video geolocalization aims to determine the precise GPS coordinates of a video's origin and map its trajectory; with applications in forensi

VPTracker: Global Vision-Language Tracking via Visual Prompt

arXiv:2512.22799v2 Announce Type: replace Abstract: Vision-Language Tracking aims to continuously localize objects described by a visual template and a language description. Existing methods, however,

WebFactory: Automated Compression of Foundational Language Intelligence into Grounded Web Agents

arXiv:2603.05044v2 Announce Type: replace Abstract: Current paradigms for training GUI agents are fundamentally limited by a reliance on either unsafe, non-reproducible live web interactions or costly

A Benchmark for Gap and Overlap Analysis as a Test of KG Task Readiness

arXiv:2604.10853v1 Announce Type: new Abstract: Task-oriented evaluation of knowledge graph (KG) quality increasingly asks whether an ontology-based representation can answer the competency questions

A Compact and Efficient 1.251 Million Parameter Machine Learning CNN Model PD36-C for Plant Disease Detection: A Case Study

arXiv:2604.11332v1 Announce Type: cross Abstract: Deep learning has markedly advanced image based plant disease diagnosis as improved hardware and dataset quality have enabled increasingly accurate ne

A Metamorphic Testing Perspective on Knowledge Distillation for Language Models of Code: Does the Student Deeply Mimic the Teacher?

arXiv:2511.05476v4 Announce Type: replace-cross Abstract: Transformer-based language models of code have achieved state-of-the-art performance across a wide range of software analytics tasks, but thei

A molecular clock for writing systems reveals the quantitative impact of imperial power on cultural evolution

arXiv:2604.10957v1 Announce Type: cross Abstract: Writing systems are cultural replicators whose evolution has never been studied quantitatively at global scale. We compile the Global Script Database

A Proposed Biomedical Data Policy Framework to Reduce Fragmentation, Improve Quality, and Incentivize Sharing in Indian Healthcare in the era of Artificial Intelligence and Digital Health

arXiv:2604.11125v1 Announce Type: new Abstract: India generates vast biomedical data through postgraduate research, government hospital services and audits, government schemes, private hospitals and t

A Quantitative Definition of Intelligence

arXiv:2604.10873v1 Announce Type: new Abstract: We propose an operational, quantitative definition of intelligence for arbitrary physical systems. The intelligence density of a system is the ratio of

A Triadic Suffix Tokenization Scheme for Numerical Reasoning

arXiv:2604.11582v1 Announce Type: cross Abstract: Standard subword tokenization methods fragment numbers inconsistently, causing large language models (LLMs) to lose positional and decimal structure -

A Two-Stage Dual-Modality Model for Facial Emotional Expression Recognition

arXiv:2603.12221v2 Announce Type: replace Abstract: This paper addresses the expression (EXPR) recognition challenge in the 10th Affective Behavior Analysis in-the-Wild (ABAW) workshop and competition

Achieving Skilled and Reliable Daily Probabilistic Forecasts of Wind Power at Subseasonal-to-Seasonal Timescales over France

arXiv:2511.16164v2 Announce Type: replace Abstract: In a growing renewable based energy system, accurate and reliable wind power forecasts are crucial for grid stability, balancing supply and demand a

Advancing Polish Language Modeling through Tokenizer Optimization in the Bielik v3 7B and 11B Series

arXiv:2604.10799v1 Announce Type: cross Abstract: The development of the Bielik v3 PL series, encompassing both the 7B and 11B parameter variants, represents a significant milestone in the field of la

AEG: A Baremetal Framework for AI Acceleration via Direct Hardware Access in Heterogeneous Accelerators

arXiv:2604.09565v1 Announce Type: cross Abstract: This paper introduces a unified, hardware-independent baremetal runtime architecture designed to enable high-performance machine learning (ML) inferen

Ambivalence/Hesitancy Recognition in Videos for Personalized Digital Health Interventions

arXiv:2604.11730v1 Announce Type: new Abstract: Using behavioural science, health interventions focus on behaviour change by providing a framework to help patients acquire and maintain healthy habits

ANCHOR: Branch-Point Data Generation for GUI Agents

arXiv:2602.07153v2 Announce Type: replace Abstract: End-to-end GUI agents for real desktop environments require large amounts of high-quality interaction data, yet collecting human demonstrations is e

Are Pretrained Image Matchers Good Enough for SAR-Optical Satellite Registration?

arXiv:2604.10217v1 Announce Type: new Abstract: Cross-modal optical-SAR (Synthetic Aperture Radar) registration is a bottleneck for disaster-response via remote sensing, yet modern image matchers are

AtlasKV: Augmenting LLMs with Billion-Scale Knowledge Graphs in 20GB VRAM

arXiv:2510.17934v2 Announce Type: replace-cross Abstract: Retrieval-augmented generation (RAG) has shown some success in augmenting large language models (LLMs) with external knowledge. However, as a

local-aiarxiv-cs-ai

Automating Structural Analysis Across Multiple Software Platforms Using Large Language Models

arXiv:2604.09866v1 Announce Type: cross Abstract: Recent advances in large language models (LLMs) have shown the promise to significantly accelerate the workflow by automating structural modeling and

BadGraph: A Backdoor Attack Against Latent Diffusion Model for Text-Guided Graph Generation

model-releasesarxiv-cs-cl

arXiv:2510.20792v4 Announce Type: replace-cross Abstract: The rapid progress of graph generation has raised new security concerns, particularly regarding backdoor vulnerabilities. While prior work has

Beyond RAG for Agent Memory: Retrieval by Decoupling and Aggregation

arXiv:2602.02007v3 Announce Type: replace-cross Abstract: Agent memory systems often adopt the standard Retrieval-Augmented Generation (RAG) pipeline, yet its underlying assumptions differ in this set

Beyond Reconstruction: Reconstruction-to-Vector Diffusion for Hyperspectral Anomaly Detection

arXiv:2604.11390v1 Announce Type: new Abstract: While Hyperspectral Anomaly Detection (HAD) excels at identifying sparse targets in complex scenes, existing models remain trapped in a scalar 'reconstr

safetyarxiv-cs-cv

Beyond Statistical Co-occurrence: Unlocking Intrinsic Semantics for Tabular Data Clustering

arXiv:2604.10865v1 Announce Type: new Abstract: Deep Clustering (DC) has emerged as a powerful tool for tabular data analysis in real-world domains like finance and healthcare. However, most existing

Beyond the Beep: Scalable Collision Anticipation and Real-Time Explainability with BADAS-2.0

model-releasesarxiv-cs-cl

arXiv:2604.05767v2 Announce Type: replace-cross Abstract: We present BADAS-2.0, the second generation of our collision anticipation system, building on BADAS-1.0, which showed that fine-tuning V-JEPA2

Applications

Bootstrapping Video Semantic Segmentation Model via Distillation-assisted Test-Time Adaptation

arXiv:2604.10950v1 Announce Type: new Abstract: Fully supervised Video Semantic Segmentation (VSS) relies heavily on densely annotated video data, limiting practical applicability. Alternatively, appl

applicationsarxiv-cs-cv

Bottleneck Tokens for Unified Multimodal Retrieval

arXiv:2604.11095v1 Announce Type: cross Abstract: Adapting decoder-only multimodal large language models (MLLMs) for unified multimodal retrieval faces two structural gaps. First, existing methods rel

Budget-Aware Uncertainty for Radiotherapy Segmentation QA Using nnU-Net

arXiv:2604.11798v1 Announce Type: cross Abstract: Accurate delineation of the Clinical Target Volume (CTV) is essential for radiotherapy planning, yet remains time-consuming and difficult to assess, e

CAGenMol: Condition-Aware Diffusion Language Model for Goal-Directed Molecular Generation

arXiv:2604.11483v1 Announce Type: new Abstract: Goal-directed molecular generation requires satisfying heterogeneous constraints such as protein--ligand compatibility and multi-objective drug-like pro

safetyarxiv-cs-lg

Hardware

Can LLMs Reason About Attention? Towards Zero-Shot Analysis of Multimodal Classroom Behavior

arXiv:2604.03401v2 Announce Type: replace-cross Abstract: Understanding student engagement usually requires time-consuming manual observation or invasive recording that raises privacy concerns. We pre

hardwarearxiv-cs-ai

CARE-ECG: Causal Agent-based Reasoning for Explainable and Counterfactual ECG Interpretation

model-releasesarxiv-cs-lg

arXiv:2604.10420v1 Announce Type: new Abstract: Large language models (LLMs) enable waveform-to-text ECG interpretation and interactive clinical questioning, yet most ECG-LLM systems still rely on wea